Ischemic mitral regurgitation (IMR) is a consequence of adverse left ventricular (LV) remodeling after myocardial infarction (MI). As a result of a lack of conclusive data regarding the best surgical approach (valve repair vs. replacement) the Cardiothoracic Surgical Trials Network (CTSN) conducted two multicenter, randomized trials to evaluate the relative benefits of these two surgical approaches to IMR. Unfortunately, the CTSN IMR trials did not establish the optimal surgical approach. Results of the CTSN Severe IMR trial demonstrated no difference in LV reverse remodeling between repair and replacement groups. However, subgroup analysis highlighted the negative implications of recurrent IMR. IMR recurred much more frequently in the repair group, resulting in more heart-failure related adverse events. Importantly, repair patients with recurrent IMR had no reduction in LV volume, while repair patients without recurrence experienced LV volume reduction that was superior to patients having valve replacement. These results strongly suggest that a patient- specific approach to surgical treatment guided by preoperative imaging-based risk stratification that is predictive of recurrent IMR would be useful for optimizing surgical results. During the initial funding period of this project, our group at the University of Pennsylvania (Penn) demonstrated that measures of mitral leaflet tethering derived from pre-operative 3D echocardiography (3DE) and a custom valve modeling algorithm accurately predicted the recurrence of IMR after valve repair. The goal of this competitive renewal is to provide conclusive evidence that pre- operative risk-based repair/replacement stratification using 3DE significantly reduces recurrent IMR and, more importantly, improves LV remodeling, long-term clinical outcomes and survival for patients with IMR. We propose to use two existing data sets to achieve our intended goal expeditiously and at limited expense: (1) as part of the initially funded project we have recruited 85 patients with IMR that have had pre-repair 3DE and have been followed prospectively to assess for recurrence of IMR. We propose to continue this recruitment at Penn to enlarge our cohort to 120 patients to allow further development and validation of an optimal predictive algorithm for recurrent IMR after MV repair; (2) the CTSN IMR trials data base which includes 551 IMR patients randomized to either MV repair (n=276), MV replacement (n=125) or CABG alone (n=150); 180 of the CTSN cohort have had pre-operative 3DE. All CTSN patients also have extensive echocardiographic and long-term clinical follow-up, which is ongoing. In Aim 1 we will establish the optimal 3DE-based predictive algorithm for recurrent IMR from candidate algorithms developed from continued recruitment of IMR repair patients at Penn. In Aim 2 we will assess the benefit of using the ideal predictive algorithm from Aim 1 on the incidence of recurrent IMR and long-term clinical outcomes in the CTSN IMR Trials data base. Finally, in Aim 3 we will develop a technique for automatic 3D segmentation and geometric modeling of the mitral valve and LV to allow for real-time risk-based repair/replacement stratification in the operating room for patients having surgery for IMR.